I. Introduction
Since its inception with the fundamental work of Smith et al. [1] and Moutarlier and Chatila [2], roboticists have been trying to address scalability issues associated with an Extended Kalman Filter (EKF) based approach to SLAM. While this approach is often considered the “standard” [3] and is attractive in its simplicity (because it only requires tracking the first and second moments of the joint landmark-robot distribution), a well known fact is that EKF SLAM inference requires quadratic complexity in the number of landmarks per update to maintain the joint-posterior correlations. As a consequence, the direct application of the EKF to SLAM is limited to relatively small environments (e.g., less than 100 landmarks).